This paper has three aims. First, we show that a method of investigating birth interval dynamics proposed by Brass and Juárez more than 25 years ago adequately controls for selection effects and does not introduce more bias relative to regression models fitted to data from 66 Demographic and Health Surveys conducted in 25 African countries during the last twenty years. Almost identical results are produced by the two approaches. Second, we use the results to understand better the trajectory of birth intervals in sub-Saharan Africa. Birth intervals have lengthened in every country examined. Third, we uncover a distinctive and previously undocumented pattern of childbearing that is prevalent across sub-Saharan Africa. After allowing for time trends in birth interval length, the lengthening of birth intervals in almost every country varies little by womens age or parity. This finding offers empirical support for the view, originally proposed by the Caldwells, that the decline in fertility in sub-Saharan Africa will follow a different pattern from that elsewhere. Implications for the future trajectory of fertility declines in Africa are discussed